49 



hydrologic zones using a number of independent variables. Tbe indepen- 

 dent variables in this analysis are the average importance values for 

 each tree cluster and vine cluster in each zone. In cases where only 

 one species in a cluster was present in a zone, that importance value 

 alone was used. Table 10 lists the data set upon which this analysis 

 was based. 



Three analyses were performed using DFA. The first analysis 

 developed a classification model using only the importance value from 

 tree clusters. This model correctly classified 47 of 55 sites for an 

 overall misclassif ication rate of 14.6 percent (Table 11). In all 

 cases, errors involved an assignment to a neighboring zone. For 

 instance, a single Zone 3 site was assigned to Zone 2, and three Zone 6 

 sites were assigned to Zone 5. The greatest error in classification 

 occurred for Zone 4 sites, which is expected because of the greater 

 tolerance of species for this zone. As previously shown (Figure 4), 

 Zone 4 is that portion of the hydrologic gradient that has species com- 

 mon to all three distribution patterns identified. Three Zone 6 sites 

 were assigned to Zone 5 primarily due to the lack of Zone 6 species 

 being identified as commonly occurring. A better agreement would be 

 expected if all species were used. 



The second model examined the ability of vine cluster importance 

 values to predict zone. Although vines did not cluster as well as 

 trees, a DFA was examined because their cluster patterns were similar. 

 However, this model correctly classified only 26 of 55 zones for a 

 52 percent misclassif ication. This model was not considered adequate, 

 and the results of its classification are not presented. 



